Abandoned cart emails: how to build a sequence and measure additional sales

Someone added a product to their cart but did not complete the purchase. They may have been distracted, compared alternatives, encountered an error or found the delivery terms unsuitable. An email can help them return only if it reflects the actual situation and reaches an appropriate recipient.

The sequence should remind people of their selection, address a specific obstacle and stop after a purchase. Evaluate it through orders and its economic impact, rather than opens alone.

Adding a product to a cart and starting checkout are different events. Choose the event your store reliably sends to the email platform.

To launch the sequence, you need:

• an identified contact who can receive these messages;
• an event containing the cart contents and action timestamp;
• a delay after which an unfinished purchase warrants a reminder;
• up-to-date data on completed orders;
• a working link back to the purchase.

An address entered during checkout does not automatically permit every type of marketing email. Before launch, establish the subscription and contact usage conditions that apply in your market.

If the store cannot associate a cart with an appropriate recipient, it cannot send the email correctly. If purchase events are missing, the system may continue sending reminders to someone who has already bought.

Do not check eligibility only when someone enters the sequence. Between the first and second messages, they may buy, unsubscribe or contact support.

Before each step, check:

• whether an order has been placed;
• whether subscription status has changed;
• whether products, prices and terms are still current;
• whether another system sends a similar reminder;
• whether a support request needs a staff response;
• whether the same contact has entered an overlapping sequence again.

For example, Klaviyo recommends excluding people who placed an order after entering the flow from subsequent emails and checking purchase event synchronisation. The logic is explained in its abandoned cart email setup guide.

First email: a reminder. Show the selected products and a clear button to return. Provide a way to ask for help if checkout caused a problem.

Second email: help with the decision. Explain terms that may affect the purchase: delivery, what is included, size selection or payment methods. Use real information, rather than a generic collection of promises.

Third email: the final reminder. Add it if it offers useful new information. Any additional offer must fit the store’s economics. After a limited sequence, stop reminders about that cart.

Do not create false urgency. Mention an expiring promotion or limited stock only when this is actually confirmed.

Suppose the store uses the checkout-start event and has the required data and basis for sending these emails.

This is an example for agreeing the logic, not a universal schedule. Intervals may differ for expensive products, quick purchases and seasonal offers. The timing of all three steps here is measured from the original event.

Do not automatically add a discount to the first email. A reminder or an explanation of delivery may be enough. Routinely offering a discount after unfinished checkout can teach customers to wait for it.

If testing a discount, include it in your margin calculation and compare it with a version without a discount. Additional orders do not always produce additional profit.

A platform may attribute an order to an email under its own rules, for example after an open or click within a defined period. This report is useful, but it does not prove the purchase would not have happened without the email.

To estimate incremental impact, use a control group that does not receive the sequence being studied. This differs from comparing two email variants; the distinction is explained in Klaviyo’s holdout group documentation. Built-in tools may have access restrictions and a different testing scope, so check exactly what they measure.

When testing a specific sequence:

1. Randomly assign eligible contacts to the groups.
2. Keep each person in the same group so another abandoned cart does not move them between groups.
3. Change only whether they receive the sequence being studied; keep other conditions comparable.
4. Define the observation period and what counts as a completed order in advance.
5. Count purchases in both groups regardless of whether the customer clicked an email.
6. Compare the share of buyers, revenue after refunds and contribution after variable costs.

Do not disable essential messages about existing orders for the control group.

In this hypothetical example, the difference is 2 percentage points, or 20 additional buyers per 1,000 contacts.

Suppose each buyer placed one order, average revenue after discounts and refunds was €80, and 35% remained after product costs, fees and other variable expenses.

The estimated increase in revenue is then 20 × €80 = €1,600, and contribution before email costs is €1,600 × 35% = €560. If the sequence being studied cost €200 during that period, €360 remains before other fixed expenses and taxes.

All figures are hypothetical. In a real test, consider whether there is enough data and account for random variation. If order values and repeat purchases differ, compare actual revenue and contribution per contact rather than a single average order value.

• Cart and purchase events arrive without critical delays.
• The link restores the correct products and works on a phone.
• Purchases and unsubscribes stop further marketing reminders.
• Prices, availability and delivery information are current.
• Repeat entry and overlapping sequences do not create duplicates.
• Recipient replies reach a member of staff.
• Reporting separates delivery and clicks from orders, refunds and economic impact.

If people repeatedly abandon checkout at the same step, check the store itself. An email will not fix broken payments or unclear delivery terms.

You can discuss sequence preparation, event integration and performance monitoring as part of our email marketing and automation service. Launch and ongoing management terms are set out in our commercial proposal.